Mast Cell Tryptase Reduces Junctional Adhesion Molecule-A (JAM-A) Expression in Intestinal Epithelial Cells: Implications for the Mechanisms of Barrier Dysfunction in Irritable Bowel Syndrome

OBJECTIVES: The objective of this study was to investigate how mast cell tryptase may influence intestinal permeability and tight junction (TJ) proteins in vitro and explore translation to irritable bowel syndrome (IBS).

METHODS: We investigated the effect of: (1) tryptase on Caco-2 monolayers, (2) mast cell degranulation in a Caco-2/ human mast cell-1 (HMC-1) co-culture model, (3) mast cell degranulation±tryptase inhibition with nafamostat mesilate (NM). Epithelial integrity was assessed by transepithelial resistance (TER), perme- ability to fluorescein isothiocyanate (FITC)-dextran and transmission electron microscopy (TEM). The expression of junctional proteins zonula occludens-1 (ZO-1), junctional adhesion molecule-A (JAM-A), claudin-1 (CLD-1), CLD-2, CLD-3, occludin and E-cadherin was determined by western blot analysis and immunofluorescence confocal microscopy. Based on the in vitro results, we further assessed JAM-A expression in biopsy tissue (cecum) from 34 IBS patients, 12 controls, and 8 inflammatory controls using immunofluorescence confocal microscopy and explored associations between JAM-A and IBS symptoms.

RESULTS: Tryptase disrupted epithelial integrity in Caco-2 monolayers as shown by a significant decrease in TER, an increase in permeability to FITC-dextran, and a decrease in the expression of junctional proteins JAM-A, CLD-1, and ZO-1 within 24 h. Correspondingly, in the Caco-2/HMC-1 co-culture model we showed a significant decrease in TER, an increase in permeability to FITC-dextran, and the presence of open TJs (TEM) in response to mast cell degranulation within 24 h. In this co-culture model, mast cell degranulation significantly decreased JAM-A and CLD-1 protein expression at 24 h. Tryptase inhibition (NM) significantly reduced the effect of mast cell degranulation on the junctional protein JAM-A, TER, and FITC-dextran flux. In IBS, epithelial JAM-A protein expression was signifi- cantly reduced in IBS tissue compared with controls. Lower JAM-A expression was associated with more severe abdominal pain (rs = − 0.69, P = 0.018) and longer duration of symptoms (rs = − 0.7,
P = 0.012) in IBS-alternating subtype.

CONCLUSIONS: Reduced JAM-A expression in vitro appears to contribute to the underlying mechanisms of altered epithelial integrity in response to tryptase released from degranulating mast cells. In IBS, JAM-A expression was significantly reduced in the cecal epithelium and associated with abdominal pain severity. JAM-A may provide new insights into the underlying mechanisms in IBS.

INTRODUCTION

Irritable bowel syndrome (IBS) is a highly prevalent functional bowel disorder characterized by abdominal pain or discomfort and associated with altered bowel habit, abdominal bloating, and disturbed defecation (1). The pathogenesis of IBS remains unclear, but is believed to result from a complex interaction between several factors including visceral hypersensitivity, psychological factors, brain–gut interactions, and immune activation. Emerging factors such as altered intestinal permeability have been added to this list (2–4).

Intestinal epithelial permeability is regulated by a complex pro- tein system comprising tight junction (TJ) and adherens junction proteins (5). These proteins include TJ proteins such as claudins (CLDs), occludin, the zonula occludens (ZO), junctional adhesion molecule (JAM), and the adherens junctional protein E-cadherin. A number of studies have suggested an increase in intestinal per- meability (2–4,6) in IBS patients. Recently, a decrease in ZO-1, occludin and claudin-1 (CLD-1) was identified in colonic biop- sies from IBS patients (7). Reduced ZO-1 expression was also reported by Piche et al. (2) in epithelial cells incubated with tissue supernatants from IBS patients. The role of E-cadherin and JAM in barrier function in IBS remains largely unexplored. E-cadherin may be associated with risk of developing post-infectious IBS (8). Although JAM-A regulates intestinal epithelial permeability (9,10) and is reduced at sites of active inflammation in inflammatory bowel disease (IBD) (11), its role in IBS is unknown.

Increased mast cell numbers in the colonic mucosa of IBS patients (12–17) is generally well documented, as is their inter- action with enteric nerves and association with abdominal pain, although not all studies support these findings (18). Beyond brain– gut interactions, the involvement of mast cells in other pathogenic mechanisms, such as intestinal permeability, are incompletely understood. We proposed that mast cells, and specifically the mast cell protease tryptase, may regulate intestinal permeability in IBS. Santos et al. (19) have shown that chronic stress resulted in increased mast cell numbers and activation in parallel with epi- thelial barrier dysfunction in animal models. More recent data suggest that tryptase may increase intestinal permeability (20) and downregulate ZO-1 expression in vitro (21).

Our aim was to investigate mechanisms by which the increased mast cell activity documented in IBS may alter intestinal perme- ability and TJ proteins. Based on emerging data, we hypothesized that tryptase released from mast cells would reduce colonic epi- thelial integrity and alter the expression of junctional proteins. We developed a human in vitro epithelial-mast cell model to inves- tigate the effects of mast cell tryptase on epithelial integrity and on the expression of the junctional proteins JAM-A, ZO-1, CLDs, occludin, and E-cadherin. Finally, we aimed to translate the key in vitro junctional protein findings to IBS. To achieve this, we inves- tigated JAM-A protein levels in IBS tissue and its potential clini- cal relevance as determined by associations with IBS symptoms, namely, abdominal pain, diarrhea, and duration of symptoms.

METHODS

Material and reagents

Reagents were purchased from Sigma-Aldrich, Arklow, Ireland unless stated otherwise. Mouse anti-ZO-1, rabbit anti-CLD-1, rab- bit anti-CLD-3, mouse anti-CLD-2, rabbit anti-occludin, rabbit anti-JAM-A, goatanti-rabbit Alexa 633-conjugatedantibodieswere from Zymed Laboratories (Paisley, UK). Mouse anti-E-cadherin was sourced from BD Biosciences (Oxford, UK) and mouse anti-glyceraldehyde-3-phosphate dehydrogenase from Millipore (Cork, Ireland). Horseradish peroxidase-conjugated anti-mouse and horseradish peroxidase-conjugated anti-rabbit antibodies were purchased from Pierce (Dublin, Ireland) and BD Pharmingen (Oxford, UK), respectively. Fluorescein isothiocyanate (FITC)- conjugated goat anti-mouse and anti-rabbit antibodies were pur- chased from Jackson ImmunoResearch (West Grove, PA).

Cell culture

Caco-2 cells. Colonic human epithelial cell line, Caco-2, was pur- chased from ECACC, Salisbury, UK. Cells (P13 to P35) were cultured in Dulbecco’s modified Eagle’s medium with 2 mM L-glutamine, 1% v/v non-essential amino acids, 10% v/v fetal bovine serum. Cells were cultured in 75 cm2 tissue culture flasks until confluent, then seeded at a density of 5×105 cells/ml on Transwell® polyester filters (0.4 m pore), and cultured until established differentiated and polarized monolayer, from 21 to 26 days, at 37 °C in a humidified atmosphere with 5% CO2 with feeding on alternate days.

HMC-1 cells. The human mast cell line, HMC-1, was a gift from J.H. Butterfield, Mayo Clinic, Rochester, MN. HMC-1 cells (P68– P94) were cultured in Dulbecco’s modified Eagle’s medium with 2 mM L-glutamine, 1% v/v non-essential amino acids, 10% iron supplemented calf serum, 1.2 mM alpha-thioglycerol. Cells were cultivated in 75 cm2 tissue culture flasks and passaged approxi- mately once per week (37 °C, 5% CO2).

Co-culture of Caco-2 and HMC-1 cells. Caco-2 cells (P22–P35) were seeded at a density 5×105 cells/ml on Transwell® filters. HMC-1 cells (P77–P95, 5×105 cells/ml) were added to the basola- teral compartment of the Transwells. Cells were co-cultured until Caco-2 established differentiated and polarized monolayer, from 21 to 23 days, in Dulbecco’s modified Eagle’s medium with 2 mM L-glutamine, 1% v/v non-essential amino acids, 5% fetal bovine serum, 5% calf serum with feeding on alternate days (37 °C, 5% CO2). In parallel, Caco-2 cells were also cultured without HMC-1 cells as controls for the co-culture.

Integrity and paracellular permeability of Caco-2 monolayers incubated with tryptase
The integrity of Caco-2 monolayers was determined with transep- ithelial resistance (TER), measured with an EVOM voltohmmeter (World Precision Instruments, Sarasota, FL). Polarized monol- ayers of Caco-2 cells were washed with pre-warmed Hank’s bal- anced salt solution (HBSS) buffer containing 11 mM glucose and 25 mM HEPES (HBSS/HEPES). Tryptase (3 or 15 mU) was added to the apical compartment of the Transwells and incubated for up to 24 h. TER was measured periodically over 24 h and the results were displayed as a percentage change over the untreated control. To assess epithelial permeability, FITC-dextran (4 kDa, 100 g/ml) was added to the apical compartment of the Transwells up to 4 h. Basolateral samplings were taken at intervals and were replenished with fresh pre-warmed HBSS/HEPES buffer at each sample time- point. The apical-to-basolateral flux of FITC-dextran was measured with a Thermo Fisher Varioskan Flash spectrophotometer (Thermo Fisher Scientific, Loughborough, UK) using an exter- nal standard curve. The apparent permeability coefficient (Papp) of each treatment can be calculated according to Papp = dQ/dt (1/AC0), where dQ/dt is the permeability rate derived from the slope of the line, A is the diffusion area and C0 is the initial donor solution concentration (22).

Development of an epithelial-mast cells co-culture model

Caco-2 cells were seeded on filters and HMC-1 cells were added to the basolateral compartment of the Transwells either at 1st, 15th, or 18th day of culture. The integrity of Caco-2 monolayers was determined with TER on alternate days, the morphology of both cell lines was analyzed with a light microscope, and the viability of HMC-1 was assessed periodically with Trypan blue staining. HMC-1 cells (5–6×106 cells/ml) were degranulated with com- pound 48/80 (5 g/ml) in Dulbecco’s modified Eagle’s medium or HBSS/HEPES and TER and FITC-dextran flux were measured over 24 h as outlined previously. The results were displayed as a change over the undegranulated control.
The effect of nafamostat mesilate (NM), a tryptase-specific inhibi- tor when used at low concentrations (10− 10–10− 11 M) (23,24), on TER and permeability was also examined by incubating cell co-cultures with the NM for up to 24 h together with mast cell degranulation. To examine the ultrastructure of the co-culture model, polarized monol- ayers of Caco-2 cells or Caco-2/HMC-1 monolayers were rinsed with warm HBSS/HEPES buffer and fixed in HBSS with 4% glutaralde- hyde. Cells were processed into resin, ultramicrotomy, and contrasted in preparation for examination by transmission electron microscopy (TEM) with a Tecnai 12 microscope (FEI, Hillsboro, OR).

Expression of TJ/adherens junction proteins by western blotting Polarized monolayers of Caco-2 cells or Caco-2/HMC-1 co- cultures were lysed in cold RIPA buffer with protease inhibitor cocktail (Roche Applied Science, Burgess Hill, UK) for 20 min on ice and sonicated for 1 min. Protein content was quantitated and equal amounts (20 g) were separated by sodium dodecyl sulfate– polyacrylamide gel electrophoresis (6–10% acrylamide, depend- ing on the protein) and transferred to nitrocellulose membrane (Whatmann, VWR International, Dublin, Ireland). Membranes were blocked in Tris-buffered saline with either 5% nonfat dry milk, 0.1% bovine serum albumin (BSA), 0.1% Tween 20 (CLD-1, CLD-3, and E-cadherin) or 5% BSA and 0.1% Tween 20 (JAM-A, ZO-1, and CLD-2) or 10% non-fat dry milk, 0.1% BSA, 0.1% Tween 20 (occludin) or 5% nonfat dry milk and 0.1% Tween 20 (glyceraldehyde-3-phosphate dehydrogenase). Blots were incu- bated with a primary antibody (0.05–1 g/ml) overnight at 4 °C and were washed three times with Tris-buffered saline containing 1% Tween-20 and incubated with horseradish peroxidase-conjugated goat anti-mouse or anti-rabbit antibodies (1:3,000–50,000) as appropriate for 1 h at room temperature. Proteins were detected by chemiluminescence (Millipore). The density of each individual band was compared with the corresponding control band and normalized against glyceraldehyde-3-phosphate dehydrogenase (loading control protein) by densitometry. ImageJ software was used to analyze western blot signals and to adjust contrast and brightness of the images (http://rsbweb.nih.gov/ij/). The results were expressed as a change relative to the untreated control.

Expression of TJ/adherens junction proteins by immunofluorescence confocal microscopy

Polarized monolayers of Caco-2 cells were rinsed with pre- warmed PBS and permeabilized with cold methanol (− 20 °C) for 30 min. Nonspecific binding sites were blocked with 1% BSA in PBS for 10 min. Cells were incubated in 1% BSA in PBS with pri- mary antibodies as follows: anti-JAM-A, anti-ZO-1, anti-CLD-1, anti-CLD-3, anti-occludin (5–20 g/ml) for 1 h at room tem- perature. Cells were washed three times with 1% BSA in PBS and incubated with FITC-conjugated goat-anti mouse or anti-rabbit antibodies (1:50–100) as appropriate for 1 h at room temperature. Cells were washed three times with 1% BSA in PBS and postfixed with 4% paraformaldehyde for 10 min. Monolayers were mounted on slides with vectashield containing 4,6-diamidino-2-phenylindole (Vector Laboratories, Burlingame, CA), and covered with a cover- slip. Specimens were stored at 4 °C before analysis on an Olympus Fluoview confocal FV100 microscope (Hamburg, Germany).

Expression of JAM-A in IBS tissue

JAM-A was immunostained in formalin-fixed, paraffin-embedded biopsy tissue from the cecum of patients with IBS (n = 34), nor- mal controls (n = 12), and inflammatory controls with confirmed IBD (n = 8) (25). Ethical approval was received from the Adelaide and Meath Hospital Ethics Committee and informed consent was obtained from participants. Patients had Rome II criteria IBS (26) macroscopically and histologically normal colonic mucosa and no evidence of organic bowel disease. Control patients were undergo- ing colonoscopy for clinical reasons (colon cancer screening, hem- orrhoids, anemia, and vomiting) and were free from organic bowel disease and IBS. Other exclusion criteria applied to IBS and normal controls were current use of the following medications: non-steroidal anti-inflammatory drugs, corticosteroids, aspirin, anti-inflammatory drugs, mast cell stabilizers, or antibiotics. Severity of abdominal pain was assessed in IBS using a pain rating of 0–4, where patients rated pain as: 0, absent; 1, mild (not influencing usual activities); 2, relevant (diverting from, but not urging modification of, usual activities); 3, severe (influencing usual activities markedly enough to urge modifications); 4, extremely severe (precluding daily activi- ties) (27). Severity of diarrhea was quantified in IBS by the reported number of soft/liquid stools per week, based on a similar variable used in IBD, namely the Crohn’s Disease Activity Index (28).
Tissue sections (3–4 m) were stained using rabbit anti-JAM- A (20 g/ml) primary antibody (Dako, Ely, UK) according to the manufacturer’s instructions. Slides were visualized using confo- cal microscopy (Olympus Fluoview FV100) and assessed by the researcher blinded to the sample identity. Immunofluorescence surface epithelium staining was graded semiquantatively on a four-point scale: (1—no specific staining, 2—mild specific stain- ing, 3—moderate specific staining, and 4—strong specific stain- ing). JAM-A was graded in all fields of view that contained surface epithelium and the median grade was calculated. Representative photomicrographs showing each grade of staining for JAM-A are displayed in Supplementary Figure 1.

Statistical analysis

Statistical analysis was performed using SPSS (Armonk, NY). Results are expressed as mean and s.e.m. or as median and range. For cell culture experiments, one-way analysis of variance was used for multiple comparisons followed with the post hoc Tukey’s test whereas single comparisons were performed with two-tailed.

RESULTS

Tryptase decreases epithelial integrity and increases paracel- lular permeability in Caco-2 monolayers

Application of tryptase (15 mU) to Caco-2 cell monolayers had a dramatic effect on epithelial integrity as seen by a significant drop in TER within 1 h of incubation that was maximal at 4 h (36±11%, P = 0.008) compared with controls (Figure 1a). This effect was partially reversible after 24 h of incubation (Figure 1b). The drop in TER was associated with a twofold (P = 0.048) increase in per- meability of Caco-2 monolayers to FITC-dextran (Figure 1c,d) at 4 h compared with controls. For the lower concentration of tryp- tase (3 mU), there was a tendency toward the reduction in TER levels compared with untreated controls, however, the difference was not statistically significant. Also 3 mU tryptase did not signifi- cantly alter permeability to FITC-dextran (Figure 1).

Tryptase decreases the expression of JAM-A, CLD-1, and ZO-1 in Caco-2 monolayers
JAM-A expression was significantly decreased when Caco-2 monolayers were exposed to tryptase (15 mU) for 4 h (0.68±0.12- fold, P = 0.03) as assessed with western blotting (Figure 2a,b), which was concomitant with the drop in TER and increased para- cellular permeability. Furthermore, confocal microscopy shown that the intensity of staining for JAM-A in TJs was clearly dimin- ished, and JAM-A was redistributed towards tricellular junc- tions (Figure 2c). The effect of tryptase on JAM-A expression was significantly reversible after prolonged incubation (24 h) and was in line with the observed partially reversible effect on TER. CLD-1 was also significantly reduced (0.68±0.01-fold, P = 0.03) relative to controls after 24 h of incubation with 15 mU tryptase, although, initially elevated expression (1.97±0.35-fold, P = 0.05) was observed at 4 h as shown with western blotting (Figure 2a,b). CLD-1 intensity within TJs did not change significantly; however,additional CLD-1 staining appeared in punctate cytoplasmic areas in cells exposed to 15 mU tryptase for 4 h, which were not observed in controls. This cytoplasmic CLD-1 was not apparent following prolonged incubation (24 h, 15 mU tryptase) and, consistent with the western blot analysis, overall CLD-1 staining in TJs was lower (Figure 2c). The lower concentration of tryptase (3 mU) did not affect JAM-A or CLD-1 expression. The western blot analysis of ZO-1 showed that prolonged exposure (24 h) of Caco-2 monolayers to tryptase (3 or 15 mU) significantly decreased its levels (0.44±0.05-fold, P = 0.0004 and 0.23±0.02-fold, P < 0.0001,respectively; Figure 2a,b). ZO-1 expression appeared elevated,but not significantly, following 4-h exposure to either 3 or 15 mU tryptase (2.40±1.06-fold, P = 0.22 and 3.85±1.91-fold, P = 0.17 respectively). Consistent with western blot findings, the inten- sity of staining for ZO-1 in TJs increased after 4 h, but decreased after 24 h of incubation with 15 mU tryptase compared with controls (Figure 2c).

CLD-3 increased after 4 h of exposure to 15 mU tryptase (2.62±0.40-fold, P = 0.06), returning to basal levels within 24 h (Supplementary Figure 2a,b) and remained unchanged in response to the lower concentration of tryptase (3 mU). Occlu- din expression increased after 4 h of incubation with tryptase (either 3 mU (1.36±0.05-fold, P = 0.001) or 15 mU (2.15±0.37-fold, P = 0.02)) but there was no detectable change after 24 h of incu- bation (Supplementary Figure 2a,b). In line with the western blot results, the levels of CLD-3 and occludin in TJs increased after 4 h of incubation with tryptase (15 mU) (Supplementary Figure 2c). Tryptase did not alter CLD-2 or E-cadherin expression in Caco-2 monolayers (Supplementary Figure 2).

Mast cell degranulation decreases epithelial integrity and increases epithelial paracellular permeability in a Caco-2/HMC-1 co-culture model

We developed a co-culture model of Caco-2/HMC-1 cells to more fully investigate the role of mast cells in the disruption of epithe- lial barrier integrity. Caco-2 cells continuously co-cultured with mast cells formed a tight epithelial barrier as demonstrated by the gradual increase in TER over time, which was maximal at day 20, and was higher when compared with Caco-2 cells alone (day 22, 170%; P < 0.0001; Figure 3a).

When HMC-1 cell were challenged with compound 48/80 to initiate degranulation, Caco-2 cells showed a significant decrease in TER within 2 h of incubation (19±2%, P < 0.0001)—this effect was maintained at a comparable level up to 24 h (Figure 3b). Degranulation of mast cells also induced a significant increase in permeability to FITC-dextran after 6 h of incubation (1.22±0.08- fold, P = 0.03) with a maximal difference at 24 h (1.68±0.07- fold, P = 0.002; Figure 3c–e) demonstrating that mast cell degranulation disrupted gastrointestinal permeability. TEM anal- ysis of Caco-2/HMC-1 co-cultures confirmed that there was no difference in TJ ultrastructure between Caco-2 monolayers alone and those co-cultured with undegranulated mast cells for 21 days. Mast cell degranulation, however, clearly disrupted epithelial integrity as determined by the presence of open TJs at 6 and 24 h in TEMs (Figure 3f), which showed a lack of electronegativity in the TJ complex.

Mast cell degranulation decreases the expression of JAM-A, CLD-1, and ZO-1 in a Caco-2/HMC-1 co-culture model
.

In the Caco-2/HMC-1 model, we confirmed that JAM-A expression was significantly reduced 6 h after mast cell degranulation (0.67±0.07- fold, P = 0.01) compared with controls, and the effect was maintained up to 24 h (0.63±0.06-fold, P = 0.0008; Figure 4a,b). This decrease in JAM-A expression following degranulation was confirmed byconfocal microscopy (Figure 4c). Interestingly, the level of CLD-1 expression in Caco-2/HMC-1 co-cultures up to 6 h after degranula- tion was comparable to controls. However, 24 h after degranulation, CLD-1 expression appeared significantly reduced (0.57±0.08-fold, P < 0.0001) as demonstrated by western blotting (Figure 4a,b). These findings were also confirmed by confocal microscopy (Figure 4c). ZO-1 levels were significantly reduced after mast cell degranulation initially (at 1, 2, and 4 h), but not at 24 h. Both JAM-A and CLD- 1, showed a significant decrease at 24 h (Figure 4), and, thus, were selected as a focus for follow-up experiments.

Tryptase inhibition diminished the effect of mast cell degranulation on epithelial integrity and paracellular permeability in a Caco-2/HMC-1 co-culture model

To better understand the role of tryptase on epithelial permeabil- ity, we investigated the effect of tryptase inhibition (NM) on TER.NM did not have a significant effect on TER of Caco-2/HMC-1 co-cultures for up to 6 h following degranulation compared with controls (Figure 5a). After 24 h, however, the reduction in TER following mast cell degranulation was significantly inhibited returning to almost control levels (P = 0.046). Consistent with this, NM significantly inhibited the effect of mast cell degranulation on paracellular permeability to FITC-dextran over 24 h of incubation (1.15±0.11-fold, P = 0.016) back to the levels of the undegranu- lated co-cultures (Figure 5b–d).JAM-A expression is significantly reduced in tissue from IBS patients.

To explore the clinical translation, we set out to determine if the novel reduction in JAM-A identified in the in vitro studies could be identified in IBS. We studied 34 IBS patients, 22 (65%) were classified as diarrhea predominant (IBS-D) and 12 (35%) as alter- nating (IBS-A) subtype (26). In the IBD group, three patients had clinically confirmed ulcerative colitis (37%) while five had Crohn’s disease (63%) whereas according to disease activity status three controls (n = 1). The reduction in JAM-A was consistent in both IBS-D (median: grade 2.5, P = 0.016) and IBS-A (median: grade 3, P = 0.041) compared with controls (Figure 7a,c). JAM-A expres- sion in inflammatory controls (IBD) was significantly lower com- pared with controls (median: inflammatory controls grade 2 vs. controls grade 4, P = 0.01) in agreement with previously published data (10) (Figure 7a,b), but not different to IBS patients.

JAM-A expression is associated with IBS symptoms

JAM-A expression in surface epithelium of cecal mucosa was sig- nificantly negatively associated with abdominal pain severity in the IBS-A subgroup (rs = − 0.69, P = 0.018; Figure 8a). The low- est JAM-A expression was observed in patients reporting severe abdominal pain (pain severity score 3; JAM-A expression median of 2) and the highest in those reporting mild abdominal pain (pain severity score 1; JAM-A expression median of 4; P = 0.025, Jonckheere–Terpstra test; Figure 8a,b). No association was found with severity of diarrhea in this study (Figure 8a). Symptom onset (years) was significantly negatively associated with JAM-A expres- sion (rs = − 0.7, P = 0.012) in IBS-A subtype, with lower expression being associated with longer duration of IBS symptoms (P = 0.017, Jonckheere–Terpstra test; Figure 8a,c). The highest expression of JAM-A was noted in IBS-A patients with relatively short duration of IBS symptoms (up to 1 year).

DISCUSSION

An increase in mast cell numbers (12–17) and an increase in intes- tinal permeability (2–4,6,29) have been independently implicatedin the pathogenesis of IBS. There are emerging data that tryptase, released from mast cells, may alter intestinal epithelial permeability (20,21), however, the mechanisms involved and the potential trans- lation to IBS have not been fully elucidated. Previously it was shown that mast cell tryptase activates and cleaves protease-activated receptor-2 (30), which is expressed both on the apical and baso- lateral membrane of the intestinal epithelial cells (31). In this study,we hypothesized that tryptase released from mast cells may reduce colonic epithelial integrity and alter the expression of junctional proteins. We showed that both tryptase and degranulated mast cells significantly increased intestinal epithelial permeability, dis- rupted epithelial integrity and decreased the expression of junc- tional proteins JAM-A and CLD-1 in vitro. Inhibition of tryptase with NM (23) mitigated the effect of mast cell degranulation on

epithelial integrity and on JAM-A. This, to our knowledge, is the first reporting of altered JAM-A expression in response to mast cell tryptase. Following up the in vitro findings, we showed for the first time, significantly lower JAM-A expression in the cecal mucosa of IBS patients compared with controls, which was asso- ciated with more severe abdominal pain and longstanding symp- toms in patients with the alternating subtype of IBS.

We have demonstrated that JAM-A levels in Caco-2 monolay- ers were significantly reduced after exposure to tryptase, which was concomitant with a drop in TER and an increase in paracel- lular permeability and was reversible at 24 h. In order to explore the mechanism of epithelial barrier disruption in response to mast cell tryptase, we developed a co-culture model of epithelial and mast cells, Caco-2/HMC-1. TEM showed that the presence of intact (undegranulated) mast cells co-cultured with epithelial cells for 3 weeks did not disturb epithelial integrity; TJ integrity of the Caco-2 cells was lost only when the underlying HMC-1 cells were degranulated. We clearly demonstrated that tryptase alone, or when released from degranulating mast cells, was involved in the disruption of the epithelial barrier and that the mechanism involves a reduction in JAM-A. The disruptive effects of tryptase on barrier function in this study are in agreement with findings from in vitro studies and in IBS biopsy tissue (20,21). Interestingly, in experimental studies, the absence of JAM-A increases intesti- nal permeability and cytokine production (9). The involvement of JAM-A, however, is new and this protein has not been investigated either in the context of IBS or in response to tryptase.
Following up on the potential translation of this finding to IBS, we show for the first time, that JAM-A protein was significantly downregulated in IBS patients relative to controls. This protein was consistently reduced in both IBS-A and IBS-D subtypes,suggesting that this alteration might be a feature of IBS independ- ent of bowel habit. Importantly, this reduction in JAM-A was sig- nificantly associated with more severe abdominal pain and longer time since symptom onset in IBS-A only. There were no associa- tions between diarrhea and JAM-A expression in this study. As this is the first reporting of JAM-A in IBS, there were no published findings to directly compare with. In IBD, however, lower JAM-A protein expression has been reported (10,11) in agreement with our findings. In addition, in this study levels in IBS and IBD did not differ significantly.

Others have linked changes in junctional proteins in IBS to symptoms. Bertiaux-Vandaële et al. (7) showed a negative asso- ciation between colonic CLD-1 protein expression and severity of abdominal pain in IBS-D. Similarly, the authors noted a negative association between occludin protein and ZO-1 mRNA expres- sion and abdominal pain in IBS (7). Interestingly, Piche et al. (2) reported a significant correlation between severity of abdominal pain and paracellular permeability in colonic biopsies of IBS patients; moreover, colonic supernatants from IBS increased paracellular permeability of Caco-2 cells and the degree of this increase was significantly associated with abdominal pain severity (2). More recently, the proportion of dilated junctions in jejunal mucosa was significantly associated with IBS symptoms including abdominal pain severity, bowel movements, and stool consistency (32). However, in this study we identified a negative association between JAM-A and pain only, but not for diarrhea.
We also report an association with duration of symptoms. JAM- A levels in IBS-A appeared to be high in patients up to 1 year after symptom onset and its expression consistently decreased with duration of IBS symptoms. Others, in contrast to our find- ing, reported the lowest levels of TJs (occludin and CLD-1 protein and ZO-1 mRNA) within the first 2 years after the onset of IBS, suggesting that altered expression of TJ proteins may be impli- cated in the initiation stage of IBS (7). Our data suggest the involvement of JAM-A in more sustained and painful IBS, possi- bly secondary to other pathological mechanisms such as sustained immune activation.

Changes in CLD-1 and ZO-1 have been published in IBS (7,33). In our in vitro experiments, CLD-1 was consistently significantly decreased at 24 h both in Caco-2 monolayers in response to tryp- tase and in response to degranulated mast cells in the Caco-2- HMC-1 model and was accompanied by increased permeability. But in contrast to JAM-A, tryptase inhibition did not alter CLD-1 expression in the co-culture model—thus, suggesting that mast cell mediators other than tryptase may have a role in its regulation. Others (21) have shown no apparent change in CLD-1 in intestinal epithelial cells in responses to protease-activated receptor-2 ago- nists. In IBS, a trend for lower CLD-1 levels in the colonic mucosa (7) had been reported, but a recent study found no difference in jejunal CLD-1 between IBS-D and controls (32).

We showed that ZO-1 protein levels in Caco-2 cells were signifi- cantly decreased after 24 h of tryptase exposure and accompanied by an increase in epithelial permeability, in agreement with others (21). This finding may fit with reports of a downregulation of ZO-1 mRNA in Caco-2 cells incubated with supernatants from colonic biopsies (2) and in ZO-1 downregulation in colonic (7) and jejunal tissue from IBS patients and the significant association between ZO-1 mRNA and tryptase mRNA (33). In the Caco-2/HMC-1 model, however, initial decreases in ZO-1 were observed up to 4 h after mast cell degranulation, but this decrease was not apparent at 24 h. It may be that ZO-1 downregulation in IBS is maintained by other non-mast cell related factors.

This study focused on the effects of mast cell tryptase on epithe- lial integrity in vitro and we sought to translate the novel JAM-A finding to IBS. Further work is required to more fully understand the mechanisms underlying lower JAM-A expression in IBS. Based on the in vitro findings, we suggest that this may involve tryptase, however, the mechanisms are likely to be complex and modified by other mast cell, immune, and clinical factors. Although the poten- tial importance of JAM-A and its link with tryptase was initially identified using apically applied tryptase in Caco-2 models, the more relevant co-culture model, inhibition, and clinical studies further confirmed a role for JAM-A. This may fit with findings of increased tryptase in IBS biopsies (33,34). Further studies are required to tease out the mechanisms underlying reduced JAM-A in IBS, including its association with tryptase in biopsy tissue, with activated mast cells using TEM and with intestinal permeability. Importantly, the association between low JAM-A and more severe pain and longer disease duration documented in this study, indi- cates that this protein may be clinically relevant to IBS pathogen- esis or treatment.

In conclusion, this study provides a potential novel mechanism in IBS and demonstrates that tryptase released from mast cells disrupts epithelial integrity via a reduction in JAM-A. Tryptase inhibition significantly reduces the disturbing effect of mast cell degranulation on TER, permeability to FITC-dextran and on the expression of JAM-A protein. We confirmed that JAM-A protein was clinically relevant in IBS, showing significantly lower levels than in controls and an association with more severe abdominal pain and longer disease duration. These findings may offer insights into underlying mechanisms FUT-175 and therapeutic targets for IBS.